I have made a prediction algorithm in which I have predicted whether a patient has cancer or not based on the past data. I have also run the model successfully and have received the parameters. Now my question is, which parameter should I give the most importance for this case of prediction? Is it the precision, recall, accuracy or the threshold?
Also you can read about global surrogate models here (in general this is a great book on interpretability):
https://christophm.github.io/interpretable-ml-book/global.html
You can also get the global importances on the model directly (if you are using scikit-learn you can use feature_importances_ property, see: https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestClassifier.html#sklearn.ensemble.RandomForestClassifier.feature_importances_).
@MUJEEBURRAHMAN-1177 Thanks, If your ML model has good performance metrics (you should decide which works best based on confusion matrix and which error can be handled and which should be reduced) then it is a great solution. Can you please share link to the model and the features that you are trying, also please share what feature engineering have you tried.
Could you please add more details about how you’re measuring your model error. If it’s doing worse than validation against your test set, then something weird is going on. If it’s doing better but not reaching 100% accuracy, that’s probably fine, and maybe preferable since it suggests that you’re not overfitting.
@@MUJEEBURRAHMAN-1177 Just checking in to see any update on this issue. If this answers your query, do click “Mark as Answer” and Up-Vote for the same which might be beneficial to other community members reading this thread. And, if you have any further queries do let us know.
2 people are following this question.
